JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ???, XXXX, DOI:10.1029/2005JD006079,

A new Greenland ice core chronology for the last glacial

termination————————————————————————

S. O. Rasmussen1, K. K. Andersen1, A. M. Svensson1, J. P. Steffensen1, B. M.

Vinther1, H. B. Clausen1, M.-L. Siggaard-Andersen1,2, S. J. Johnsen1, L. B.

Larsen1, D. Dahl-Jensen1, M. Bigler1,3, R. Rothlisberger3,4, H. Fischer2, K.

Goto-Azuma5, M. E. Hansson6, and U. Ruth2

Abstract. We present a new common stratigraphic time scale for the NGRIP and GRIPice cores. The time scale covers the period 7.9–14.8 ka before present, and includes theBølling, Allerød, Younger Dryas, and Early Holocene periods. We use a combination ofnew and previously published data, the most prominent being new high resolution Con-tinuous Flow Analysis (CFA) impurity records from the NGRIP ice core. Several inves-tigators have identified and counted annual layers using a multi-parameter approach, andthe maximum counting error is estimated to be up to 2% in the Holocene part and about3% for the older parts. These counting error estimates reflect the number of annual lay-ers that were hard to interpret, but not a possible bias in the set of rules used for an-nual layer identification. As the GRIP and NGRIP ice cores are not optimal for annuallayer counting in the middle and late Holocene, the time scale is tied to a prominent vol-canic event inside the 8.2 ka cold event, recently dated in the DYE-3 ice core to 8236years before A.D. 2000 (b2k) with a maximum counting error of 47 years. The new timescale dates the Younger Dryas – Preboreal transition to 11,703 b2k, which is 100–150years older than according to the present GRIP and NGRIP time scales. The age of thetransition matches the GISP2 time scale within a few years, but viewed over the entire7.9–14.8 ka section, there are significant differences between the new time scale and theGISP2 time scale. The transition from the glacial into the Bølling interstadial is datedto 14,692 b2k. The presented time scale is a part of a new Greenland ice core chronol-ogy common to the DYE-3, GRIP and NGRIP ice cores, named the Greenland Ice CoreChronology 2005 (GICC05). The annual layer thicknesses are observed to be log-normallydistributed with good approximation, and compared to the Early Holocene, the meanaccumulation rates in the Younger Dryas and Bølling periods are found to be 47±2%and 88 ± 2%, respectively.

An edited version of this paper was published by AGU. Copyright 2006 American Geo-physical Union.Citation : Rasmussen, S. O., et al. (2006), A new Greenland ice core chronology for thelast glacial termination, J. Geophys. Res, 111, D06102, doi:10.1029/2005JD006079.

1. Introduction

A wealth of information about palaeoclimate can be ex-

tracted from polar ice cores, but the full potential of these

data can be exploited only with a reliable depth–age rela-

tion. Especially when studying the dramatic climatic transi-

1Ice and Climate, Niels Bohr Institute, University ofCopenhagen, Denmark

2Alfred-Wegener-Institute for Polar and Marine Research,Bremerhaven, Germany

3Climate and Environmental Physics, Physics Institute,University of Bern, Switzerland

4Now at British Antarctic Survey, Cambridge, UK5National Institute of Polar Research, Tokyo, Japan6Department of Physical Geography and Quaternary

Geology, Stockholm University, Sweden

Copyright 2006 by the American Geophysical Union.0148-0227/06/2005JD006079$9.00

tions of the past, accurate age estimates are of great impor-tance because the relative timing of climate changes aroundthe globe gives indications of the causes and mechanismsfor rapid climatic changes [Bond et al., 1993; Blunier et al.,1998]. Much effort has therefore been put into developingtime scales for ice cores, based either on identification andcounting of annual layers or modeling the depth–age rela-tionship [Hammer et al., 1978; Hammer, 1989]. Greenlandice cores can be dated by annual layer counting when theaccumulation rate is sufficient to resolve annual layers, andthe time scales of different ice cores can be matched and val-idated using volcanic layers and other independently datedstratigraphic markers [Clausen et al., 1997; Anklin et al.,1998]. The DYE-3 and GRIP ice cores were dated about 8ka back by counting annual layers in the stable isotope andelectrical conductivity measurement profiles [Hammer et al.,1986; Hammer, 1989; Johnsen et al., 1992]. Below this, an-nual layers were identified using GRIP chemistry data aswell. In the glacial part of GRIP, discontinuous annual layercounting was used as input for ice flow modeling [Dansgaardet al. 1993; Johnsen et al. 1995 (ss09 time scale); Johnsenet al., 2001 (ss09sea time scale)], while the GISP2 ice core

1

X - 2 RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY

1299.81

1334.04

1522.79

1624.27

EC

M +

CF

A (

Ca2+

, NH

4+ , H2O

2)1194.05

1228.24

1404.74

1492.45

1526.52

1574.80

1604.64CF

A (

7 sp

ecie

s) +

VS

EC

M

NGRIP δ18O(per mille)

GRIPdepth(m)

GRIPdata

NGRIPdata

NGRIPdepth(m)

Isotopeevent

Periodlabel

GS−1

GIS−1abcd

GIS−1e

GS−2a

Holocene I

Holocene II

YD

Allerød

Bølling

OD

Datum8.2 kaevent

−44 −39 −34

Figure 1. Overview of the data series used for the dif-ferent parts of the GRIP and NGRIP ice cores. VS refersto the visual stratigraphy data, ECM are electrical con-ductivity measurement data. The period labels are thenames used throughout this work to identify the differ-ent time periods. YD and OD refer to the Younger Dryasand Oldest Dryas, respectively. The isotope event namesrefer to those of Bjorck et al. [1998].

was dated using stratigraphic methods also in the glacial,relying primarily on the visual layers in the ice [Alley et al.,1997; Meese et al., 1997]. As discussed by Southon [2004],the different time scales of GRIP and GISP2 are up to sev-eral thousand years offset in parts of the glacial, and thereare significant differences in the Holocene as well.Drilling of the NGRIP ice core was completed successfullyin 2003 [Dahl-Jensen et al., 2002; NGRIP members, 2004].Liquid water was found when the drill penetrated the icesheet, revealing melting point temperatures at the bedrock.The melting at the base limits the age of the ice in theNGRIP ice core to be approx. 123 ky old [NGRIP members,2004]. The combination of moderate accumulation rates (19cm of ice equivalent per year in present time) and bottommelting result in a different flow pattern than that of GRIPand GISP2, and the annual layers are thus more than 5mm thick over the entire length of the NGRIP core. Thismeans that in the middle and early part of the glacial the an-nual layers are thicker than those observed in the GRIP andGISP2 ice cores [NGRIP members, 2004; Dahl-Jensen et al.,1993; Johnsen et al., 2001; Meese et al., 1997]. This fact,and the development of new high resolution impurity mea-surement techniques, makes the NGRIP ice core ideal forstratigraphic dating purposes, and has motivated the initia-tion of the Copenhagen Ice Core Dating Initiative, with theconstruction of a new stratigraphic timescale for the GRIPand NGRIP ice cores as one of the main objectives. Herewe present a stratigraphic time scale for the period 7.9–14.8ka before A.D. 2000 (b2k), using multi-parameter data fromboth the GRIP and NGRIP ice cores. The datum of thepresented time scale is a readily recognizable volcanic event

inside the characteristic 8.2 ka cold event, dated using datafrom the DYE-3 ice core. In the section 7.9 – 10.3 ka beforepresent, the time scale is based on new annual layer count-ing using existing impurity records from the GRIP ice core[Fuhrer et al., 1993; 1996; 1999]. From 10.3 ka b2k and backalso NGRIP impurity records are available, and down to theYounger Dryas – Preboreal transition (henceforth named theYDPB transition) the time scale is based on the combinedGRIP-NGRIP data set. For NGRIP, the time scale contin-ues through the Younger Dryas, Allerød and Bølling periodsback to 14.8 ka before present. The new time scale is namedthe Greenland Ice Core Chronology 2005, or GICC05.In order to be able to refer to the different sections of thepresented time scale in a short and unambiguous way, thenames Holocene I, Holocene II, Younger Dryas, Allerød,Bølling, and Oldest Dryas are used as illustrated in Fig-ure 1, although this use does not comply with the formalbio-stratigraphic definitions of the periods. The transitiondepths used to define the onset and end of the Bøllingand Younger Dryas periods are derived from deuterium ex-cess data (the deuterium excess is defined as δD – 8δ18O).In each of these transitions, the deuterium excess changesabruptly, and the change occurs prior to, or simultaneouslywith, changes in all other climate proxies. As changes in thedeuterium excess are connected to changes in the moisturesources [Jouzel and Merlivat, 1984; Johnsen et al., 1989;Taylor et al., 1997; Masson-Delmotte et al, 2005a; 2005b],the abrupt shifts in deuterium excess indicate that dramaticreorganizations of the atmospheric circulation took placeat the onset of these transitions, followed by more grad-ual changes in temperature and ice core impurity content.A discussion of the implications of these observations is be-yond the scope of this work, but it should be stressed thatthe timing of the transitions as defined by the deuteriumexcess must be expected to precede changes in e.g. temper-ature and vegetation recorded in other archives.All ages in this work are reported in calendar years rela-tive to the year A.D. 2000. Unfortunately the BP notationhas in several instances been used with reference to otheryears than the conventional A.D. 1950 when reporting icecores results. To avoid further confusion, and to underlinethe independency from e.g. radiocarbon-based dating, thenotation b2k is introduced, being both short and unambigu-ous.

2. Data

For NGRIP, Continuous Flow Analysis (CFA) data ofsoluble ions were used for identification of annual layers[Bigler, 2004; Rothlisberger et al., 2000]. The resolutionof the [NH+

4 ], [Ca2+], and conductivity series has been en-hanced as described in Rasmussen et al. [2005], by cor-recting for the effect of dispersion in the CFA system us-ing deconvolution techniques. The correction method usesthe measured smooth response to a sudden concentrationjump, obtained from calibration measurements, to estimatethe mixing strength and restores as much as possible of thehigh-frequency part of the signal taking into account thepresence of measurement noise. Moreover, CFA dust data(the number concentration of particles with diameter largerthan 1.0µm) of Ruth et al. [2003] were used. Electrical con-ductivity measurement (ECM) data representing the acidityof the ice were used [Dahl-Jensen et al., 2002]. All ECMprofiles are shown as [H+] in µequiv./kg. The visual stratig-raphy (VS) grey-scale refraction profile of Svensson et al.[2005] were also included, but as the raw VS data containmany close, thin layers representing sub-annual variations,we used a profile smoothed by applying a Gaussian filterwith s = 4 mm.The ice core drill got stuck in 1997 during the NGRIP

RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY X - 3

Table 1. Data sets used in the construction of the presented time scale.

Ice core Species Depth Sampling Estim. effectiveInterval (m) resolution (mm) resolutiona(mm)

NGRIP2 CFA: NH+

4, Ca2+, Conductivity (resolution enhanced) 1404–1607 1 10 − 15b

CFA: NO−

3, Na+, SO2−

4, Dust 1404–1607 1 15 − 25b

Visual Stratigraphy 1404–1607 < 1 ∼ 3ECM 1346–1607 1 4

NGRIP1 ECM 1195–1372 10 40

GRIP CFA: NH+

4, H2O2 1300–1624 2 ∼ 20c

CFA: Ca2+ 1300–1624 2 ∼ 50c

ECM 1300–1624 10 40

a The effective resolution is the shortest wavelength that can be identified from the data.b The resolution varies with depth due to changing experimental conditions.c Approximate values obtained from inspection of the data. Fuhrer et al. [1993] report the resolution defined as the e-folding scale

as 7, 12, and 35 mm for the NH+

4, H2O2, and Ca2+ subsystems, respectively

drilling operation and a new core had to be drilled. Thetwo cores are referred to as NGRIP1 and NGRIP2, respec-tively. Measurements have been performed on the NGRIP1core down to a depth of 1372 m, while measurements on theNGRIP2 core start at a depth of 1346 m (corresponding toapproximately 9.5 ka b2k). In the zone of overlap the meanoffset between NGRIP1 and NGRIP2 is 0.43 m, with thesame feature appearing at greater depths in the NGRIP1core than in the NGRIP2 core [Hvidberg et al., 2002]. Alldepths are NGRIP2 depths unless noted otherwise, and thusNGRIP1 data have been shifted 0.43 m to fit the NGRIP2depth scale.From the GRIP ice core, ECM data and the CFA records of[NH+

4 ], [H2O2], and [Ca2+] obtained by Fuhrer et al. [1993;1996; 1999] were used.

Table 1 lists the data series and the estimated effectiveresolution of each of the data series used. The effective res-olution is defined for each series as the shortest cycle thatcan be identified in that series.Model estimates of mean annual layer thicknesses in theNGRIP ice core [Johnsen et al., 2001; NGRIP members,2004] are above 5 cm in the Holocene, around 3 cm in theYounger Dryas, about 4 cm in Bølling and Allerød, and 2–3cm in the Oldest Dryas. The resolution of the CFA, ECM,and VS data thus allows identification of annual layers inthe ice from the Holocene and back through the transition,while the resolution of the CFA data becomes marginal be-low the transition into the Bølling interstadial. Preliminaryresults show that the CFA data quality improves at greaterdepths, and it is thus possible to use CFA data for the iden-tification of annual layers within most interstadials, wherethe accumulation is roughly twice that of the stadials. Dur-ing the cold stadials annual layer identification has to relymostly on VS and ECM.

3. Observed seasonality

Many of the impurity records obtained from Greenlandice cores exhibit annual variations [Beer et al., 1991; Whit-low et al., 1992; Fischer and Wagenbach, 1996]. The CFA,ECM, and VS data from GRIP and NGRIP are so highlyresolved that the intra-annual timing of the different speciesis clearly detected at all depths. In the relatively warmperiods (the Holocene, the Bølling and part of the Allerødperiod) the different species peak at different times of theyear. The relative timing of the species resembles that ob-served for recent times [Whitlow et al., 1992; Laj et al.,1992; Steffensen, 1988; Anklin et al., 1998; Fuhrer et al.,1993; Bory et al., 2002]. A typical annual layer is character-istic by having the sea salt dominated [Na+] peaking in late

winter. The VS record generally contains much more thanone peak per year and is not easy to interpret, but in generalthere are layers of high refraction (cloudy bands) at spring-time, coinciding with high dust content, high [Ca2+] anddips in the [H2O2] curve. Summer is characterized by highconcentration of [NH+

4 ], [NO−

3 ], and sometimes of [SO2−

4 ].In general, dips in the ECM correlate with peaks in [NH+

4 ]and [Ca2+], while the electrolytical conductivity (henceforthjust called the conductivity) is related to the total contentof ions present in the meltwater and thus contains severalpeaks per year due to the different seasonality of the indi-vidual species. In the colder periods, the differences in sea-sonality almost vanish and most series peak simultaneously,making the conductivity record useful for identification ofannual layers. The [NH+

4 ] signal does not consistently showclear annual cycles in the cold periods. In the colder peri-ods, the annual signal in VS becomes more prominent, butthe VS record still contains more than one peak per year onaverage.It should be noted that although there is an annual [H2O2]signal present in recently formed snow, this signal has beenerased by diffusion at the depths in question in this work.However, because [H2O2] is only preserved in ice with lowdust levels [Fuhrer et al., 1993], the dips in the [H2O2] curveindicate high dust content. As the GRIP [H2O2] measure-ments have significantly higher resolution than the corre-sponding [Ca2+] measurements, details obscured by the lowresolution of the [Ca2+] resolution are often resolved by the[H2O2] data.

4. Identification of Annual Layers

Identification and subsequent counting of annual layersin ice cores has been performed in various ways. The mostdirect, practically as well as conceptually, is to base theidentification on a single parameter, which is known to ex-hibit annual cycles. Langway, Jr., [1967], used visible fea-tures to establish one of the first stratigraphic time scalesfor a Greenland ice core, ranging a few hundred years back,but most often δ18O data are used where the accumulationrate is sufficiently high. The δ18O parameter is the obviouschoice because the close connection between δ18O and tem-perature makes it highly probable that the observed cyclesactually represent annual layers. Mainly using δ18O mea-surements, a time scale for the last 8 ka was constructed bycounting of annual layers in the DYE-3 ice core [Hammeret al., 1986]. Another prominent example of a single pa-rameter stratigraphic time scale is that of the Byrd ice core,which was dated some 50,000 years back in time primarilyusing ECM data containing clear annual cycles [Hammeret al., 1994]. When more parallel data series with suffi-cient resolution are available from the same segment of an

X - 4 RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY

5940b2k

5945b2k

5950b2k

5955b2k

5960b2k

1551.0 1551.2 1551.4 1551.6 1551.8 1552.0 1552.2 1552.4 1552.6 1552.8 1553.0

−240

−220

−200

−180

DYE−3 depth (m)

δD (

per

mill

e)

0

1

2

3

EC

M

Figure 2. A section of new δD data used for the revision of the DYE-3 time scale (dark grey), togetherwith the ECM data (blue) used in the construction of the former DYE-3 time scale of Hammer et al.[1986]. The annual layer markings of the revised time scale are shown by light grey vertical bars (datesare relative to A.D. 2000, denoted b2k).

ice core, it is obviously preferable to base the identificationof annual layers on all the available data [Johnsen et al.1992; Meese et al., 1997; Alley et al., 1997; Anklin et al.,1998]. This is especially true when the available data se-ries cannot be guaranteed to be pure annual signals, butcontain contributions from other processes than those gen-erating the annual pattern. For example, the concentrationof NH+

4 in the Greenland ice cores exhibits clear annualvariations in the Holocene and in the Bølling interstadial,but the annual signal is occasionally obscured by high peaksoriginating from e.g. biomass burning events. The use ofmultiple data series thus improves the quality of the timescale produced by making the identification of annual lay-ers more robust. However, multi-parameter data sets witha resolution sufficient for annual layer counting are sparse,and are seldom available from the brittle part of ice cores,where internal cracks in the ice make it virtually impossibleto obtain uncontaminated continuous measurements of theimpurities in the ice. This is one of the main reasons whymulti-parameter CFA measurements have only been carriedout below the depth of 1300 m in the GRIP core and be-low 1400 m in the NGRIP core. Because of the relativelylow accumulation rate at the NGRIP drill site, δ18O datafrom NGRIP are not optimal for identification of annuallayers, while for the GRIP core, δ18O measurements are notavailable with sufficient resolution to allow identification ofannual layers in the 4–8 ka part of the core.Due to the relatively high accumulation rate at DYE-3, sta-ble isotope ratios from the DYE-3 ice core thus remain, inthe opinion of the authors, the best ice core data available fordating the most recent 8 ka. However, at the time when theDYE-3 time scale of Hammer et al. [1986] was constructed,highly resolved stable isotope ratios had only been mea-sured continuously down to 5.9 ka b2k, and the time scalewas therefore to some degree based on interpolation and onECM measurements below this [Hammer, 1989]. However,the highly resolved DYE-3 isotope ratio profile has recentlybeen completed [Vinther et al., 2006]. Using the completeDYE-3 isotope data set together with GRIP δ18O data inthe 0–3.8 ka b2k interval and NGRIP δ18O data in the 0–1.9ka b2k interval, a new and much more robust cross-validatedtime scale for the DYE-3, GRIP, and NGRIP ice cores reach-ing just beyond the 8.2 ka cold event has been constructed[Vinther et al., 2006]. Figure 2 shows one of the new sec-tions of DYE-3 δD data from around 6 ka b2k together withthe corresponding ECM data initially used to construct theDYE-3 time scale. It is seen that the annual layers areclearly identifiable from the δD data without diffusion cor-rection, and it is apparent that the counting error is reducedsignificantly compared with the uncertainty of the previousdating. The most recent 1.9 ka have been dated with nocumulated uncertainty as the reference horizon of Vesuvius(A.D. 79) is dated accurately from historical records. In the1.9–3.8 ka b2k section the GRIP and DYE-3 records were

matched using common ECM events, and the annual layerswere identified from the combined records. The maximumcounting error is therefore very small, estimated to about0.25%. In the 3.8–8.3 ka b2k section the time scale is basedon DYE-3 stable isotope ratios, as illustrated in Figure 2. Inthe 3.8–6.9 ka b2k part, the estimated maximum countingerror is 0.5%. At the DYE-3 drill site, diffusion of the oxygenisotopes in the ice affects the annual signal when the annuallayer thickness is below 6 cm. Due to ice-flow induced thin-ning, the mean annual layer thickness is reduced to about6 cm at a depth of 1625 m (corresponding to 6.9 ka b2k).From 6.9 to 8.3 ka b2k the maximum counting error there-fore increases to 2%, because the annual signal gradually isweakened by diffusion in the ice. In this way, the new timescale reaches beyond the 8.2 ka cold event with a cumulatedmaximum counting error of about 50 years. The maximumcounting error has been estimated from the number of po-tential annual layers that were hard to interpret, and doesnot include a possible bias in the annual layer identificationprocess. The concept of maximum counting error will bediscussed further in section 5.1.4.A prominent ECM double peak is found close to the deepestpart of the δ18O-minimum of the 8.2 ka cold event [Hammeret al., 1986]. The layer is characterized by a high fluoridecontent, and can thus most likely be attributed to an Ice-landic volcano. Because of the special timing of the ECMdouble peak inside the δ18O minimum, this stratigraphichorizon can be uniquely identified in all Central Greenlandice cores, and has thus been chosen as the datum of thepresented time scale. According to the revised DYE-3 timescale [Vinther et al., 2006], the annual layer inside the ECMdouble peak has been dated to 8236 b2k with a maximumcounting error of 47 years. Any future changes in the datingof this horizon will propagate to the presented GICC05 timescale. The depths of this horizon in the Central Greenlanddeep ice cores are listed in Table 2.

Table 2. Depth of the ECM double peak inside the 8.2 kaevent in selected Greenland ice coresa

Ice core Depth (m)

NGRIP1 1228.67NGRIP2 1228.24b

GRIP 1334.04DYE-3 1691.06GISP2 1392.66

a The peak serves as the datum of the presented time scaleand is assigned the age 8236 b2k with a maximum counting errorof 47 years [Vinther et al., 2006].

b Estimated using the calculated offset of 0.43 m betweenNGRIP1 and NGRIP2 as described in section 2.

RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY X - 5

1393.5 1393.7 1393.9 1394.1 1394.3 1394.5 1394.7GRIP depth (m)

−35

−30

δ18O

(pe

r m

ille)

0

50

[NH

4 +] (ppbw)

0

10

20

[Ca2+

] (pp

bw)

0

50

100

[H2 O

2 ] (ppbw)

10−1

100E

CM

Figure 3. Example of 1.2 meter of GRIP data and annual layer markings (grey vertical bars) fromabout 8.8 ka b2k. The annual layers are identified as matching pairs of spring and summer indicators:spring is characterized by high dust content leading to peaks in [Ca2+] and dips in the [H2O2] curve,while summer is characterized by high [NH+

4 ] and corresponding minima in the ECM curve. Note thatthe ECM and [H2O2] curves are plotted on reversed scales. In this section the annual layer identificationprocedure is supported by high-resolution δ18O data, corrected for diffusion using the method of Johnsen[1977] and Johnsen et al. [2000] (raw data: thick line, diffusion corrected data: thin line).

4.1. Multi-parameter Annual Layer Counting in theGRIP Ice Core (Holocene I section)

Below 8.3 ka b2k, the resolution of the DYE-3 isotope sig-nal becomes insufficient for annual layer identification dueto flow-induced thinning of the layers, and the GRIP iso-tope signal is so severely dampened by diffusion that iden-tification of annual layers from the isotope profile alone isdubious. Instead, annual layers were identified from theCFA data set of Fuhrer et al. [1993; 1996; 1999], which hasalready provided the GRIP core with a stratigraphic timescale covering the period from 7.9 ka b2k to the YDPB tran-sition [Johnsen et al., 1992]. An initial comparison of thistime scale with the new NGRIP data (see below) indicatethat the existing GRIP time scale is missing a significantnumber of annual layers in the Holocene II section, and anew GRIP chronology has therefore been constructed usingthe [NH+

4 ], [H2O2], and [Ca2+] series obtained by Fuhreret al. [1993; 1996; 1999], ECM data, and short sections ofhigh-resolution isotope data (5 meter of data for every 50meters).In general the [Ca2+] series has fewer peaks than the [NH+

4 ]series, which at least partially arises from the fact that the[Ca2+] measurements have a significantly lower resolution(see table 1). When originally marking the annual layers inthe GRIP core, the [Ca2+] series was believed to be the mostreliable for annual layer identification, while the [NH+

4 ] se-ries was considered to contain additional peaks not relatedto the annual signal. If on the other hand all significantpeaks in the [NH+

4 ] signal are counted as years, the num-ber of annual layers increase by about 7%. The approachused here is based on the different seasonality of the seriesas described in section 3: the spring is characterized by highdust content leading to high [Ca2+] and dips in the [H2O2]curve, while the [NH+

4 ] has summer maxima and correspond-ing ECM minima. The annual layers have been defined asmatching pairs of these spring and summer indicators, whichis supported by the high resolution δ18O data where avail-able. One of these sections is shown in Figure 3. The factthat the dust-rich spring is observed in both the [Ca2+] and[H2O2] curves, while the summers are seen in both the [NH+

4 ]and ECM curves reduces the counting error significantly asmeasurement-related problems and resulting data gaps of-ten affect only one series at a time. When either the springor summer indication is weak, or when the relative timing of

the spring and summer indicators is unusual, an ”uncertainlayer mark” has been placed. From the start it was agreedbetween the investigators that the uncertain marks shouldbe regarded and counted as ”half years”, and the uncertainmarks have therefore been set with this in mind. The valid-ity of the applied criteria has been tested by marking annuallayers in the Holocene II section using GRIP data only, andsubsequently cross-validating with the NGRIP annual layersequence. Differences smaller than 1% were observed, andthe criteria used in the GRIP and NGRIP parts are there-fore considered to be consistent.In practice, the time scale was constructed by first lettingthree investigators (BMV, JPS, and SOR) independentlyplace annual layer marks. The three annual layer profileswere different in around 200 places in the 2.4 ky long sec-tion, but the total number of annual layer marks in thethree profiles agreed within 1.5%. Each point of disagree-ment was subsequently reviewed with a fourth investigator(HBC) acting as arbitrator. The resulting time scale repre-sents a compromise between the three initial versions, usinguncertain layer marks to mark points where unanimity couldnot be reached, or where either the spring or summer indi-cators are not clear. The resulting time scale contains about1.5% more annual layers than the previous counted GRIPscale in the Holocene I section [Johnsen et al., 2001].

4.2. Multi-parameter Annual Layer Counting in theNGRIP Ice Core

The NGRIP data set comprises an extensive set of mea-surements, where a clear annual signal is present in up to9 parallel data series. As an initial approach, three inves-tigators (KKA, AMS, and JPS) made independent annuallayer counts based on all available NGRIP data series. Inthe Holocene, Allerød, and Bølling periods, the investigatorsagreed within a few percent on the number of annual layermarks, and in many century-long sections they agreed onevery year, but in the Younger and Oldest Dryas discrep-ancies of up to 5% and 10%, respectively, were observed.The differences were most pronounced in the coldest peri-ods where thin annual layers begin to appear as shoulderson neighbouring peaks, and around sharp transitions. Againthe three initial time scales were reviewed with a fourth in-vestigator (SOR) acting as arbitrator, thereby producing a

X - 6 RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY

1464.1 1464.2 1464.3 1464.4 1464.5 1464.6 1464.7 1464.8 1464.9 1465NorthGRIP depth (m)

1591.3 1591.4 1591.5 1591.6 1591.7 1591.8 1591.9 1592 1592.1 1592.2 1592.3GRIP depth (m)

−41

−37

−33

δ18O

(pe

r m

ille)

0

35

70

[NH

4 +] (ppbw)

5

10

15

20

[Ca2+

] (pp

bw)

0

25

50

[H2 O

2 ] (ppbw)

10−1

100

EC

M

0

20

40

60

[NH

4+ ] (pp

bw)

0

20

40

[Ca

2+] (ppbw)

50

100

150

200

[NO

3− ] (pp

bw)

30

50

70

90

[Na

+] (ppbw)

0

100

200

300

[SO

42−] (

ppbw

)

0.5

1.5

2.5

Cond. (µS

/cm)

10−1

100

EC

M 0

20

40

Dust (part./µL)

40

60

80

100V

is. S

trat

.

Figure 4. Example of data and annual layer markings (grey vertical bars) from the early Holocene. Theupper nine panels show a 0.95 meter long section of NGRIP data, and the lower five panels show thecorresponding 1.05 m section in the GRIP data set. The annual layers are marked at the summer peaks,which are defined by high [NH+

4 ] and [NO−

3 ]. The spring is characterized by high dust mass leading topeaking [Ca2+] and dips in the [H2O2] profile, while the [Na+] peaks in late winter. The visual stratig-raphy profile does not contain clear annual layers, but contains peaks corresponding to almost everydust peak. The ECM (note the reverse logarithmic scale) anti-correlates strongly with the largest peaksin [NH+

4 ], but does not itself allow safe identification of annual layers. The lower four panels show thesame time interval in the GRIP core, from which [Ca2+] and [NH+

4 ] measurements exist [Fuhrer et al.,1993; 1996; 1999]. The similarity of the [NH+

4 ] records (and consequently also to some degree the ECMrecords) from NGRIP and GRIP allows a close stratigraphic matching of the two cores. The annual layeridentification has been based on impurity data only, but is supported by comparison with high-resolutionδ18O data that are available for a few short sections in the Holocene. The δ18O data have been correctedfor diffusion using the method of Johnsen [1977] and Johnsen et al. [2000] (raw data: thick line, diffusioncorrected data: thin line)

time scale where every annual layer marking was acceptableby all investigators. Ambiguous features and points, whereunanimity could not be reached, were marked by uncertainlayer marks.4.2.1. Holocene II Section, NGRIP Depth 1404.7–1492.45 m

In the Holocene, the series show strong and different sea-sonality as described in section 3 and illustrated in the up-

per part of Figure 4. Slightly different relative timing ofthe different species is often observed for one or two years,most often related to apparent merging of successive seasons,where e.g. winter and spring peaks or spring and summerpeaks occur simultaneously as observed in Figure 4, wherethe [Na+] winter peak and the following [Ca2+] and dustspring peaks occur at the same depth of 1464.87 m. Thefact that the different series have different seasonality makes

RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY X - 7

1502.2 1502.3 1502.4 1502.5 1502.6 1502.7Depth (m)

50100150200

[Ca2+

] (pp

bw)

50

150

250

[NO

3 −] (ppbw)

020406080

[Na+ ] (

ppbw

)

100

200

300

400

[SO

4 2−] (ppbw)

0.5

1.5

2.5

Con

d. (µ

S/c

m)

50

100

150

200

Dust (part./µL)

50

100

150

Vis

. Str

at.

Figure 5. Example of data (heavy lines), resolution enhanced data (thin lines), and annual layer mark-ings (grey bars) from the Younger Dryas. The different series peak almost simultaneously, and peaks inthe VS profile are seen to be connected closely to the annual layers. However, too many layers couldeasily be counted if the counting was based on VS data alone, emphasizing the importance of usingmultiple data series for annual layer counting. The uncertain layer (open grey bar) at 1502.39 m is apotential thin annual layer, not fully resolved by the measurements.

it highly improbable that full years are missing in the dataset due to post-depositional processes or missing precipita-tion, and also makes the identification of annual layers veryrobust. Of the 1436 annual layers marked in this section,only 23 marks, or 1.6%, were marked as uncertain layers.Of these 23 marks, 20 are placed where some series indicatethat an annual layer is present, while other series do nothave evidence of a layer. The remaining 3 uncertain layersare caused by data gaps or short sections with data qualityproblems.GRIP CFA data are also available in the Holocene II sec-tion, and the GRIP and NGRIP cores can be matched on anannual basis by first using major ECM horizons to provide alow-resolution stratigraphic matching of the two records andthen by matching the [NH+

4 ] profiles of the two cores yearby year in between the fix points. In Figure 4, the NGRIPdata (upper nine panels) are presented together with thecorresponding δ18O, [NH+

4 ], [Ca2+], [H2O2], and ECM se-ries from GRIP (lower five panels). In order to be able toasses the possible bias of the annual layer counting proce-dure, two investigators (BMV and HBC) constructed a timescale for the Holocene II period using the combined data setof CFA and ECM data from both NGRIP and GRIP in-dependently from the four-investigator NGRIP time scaledescribed above. The question of possible bias will be dis-cussed further in section 5.1.4.4.2.2. Younger Dryas Section, Depth 1492.45–1526.52 m

During cold periods like the Younger Dryas, those seriesshowing annual cycles mostly peak simultaneously. Thissupports the conclusions of Werner et al. [2000] that pointto that Central Greenland receives only little winter precip-itation under glacial conditions. Although there is roughlyone [NH+

4 ] peak per year on average, the [NH+

4 ] signal wasnot in general regarded as being reliable for annual layeridentification in the Younger Dryas. Also the ECM signalbecomes very hard to interpret in some sections. In theYounger Dryas, the expected mean annual layer thicknessderived from flow modeling is less than twice the wavelengthof the shortest cycle that can be resolved by the CFA mea-surements. Thus, thin annual layers may be poorly resolvedor in exceptional cases vanish. When identifying the an-nual layers, special consideration was put into identifying

features that could represent two almost merged layers. InFigure 5 the usable data series (original and resolution en-hanced) are shown together with the annual layer markings.The uncertain layer at depth 1502.39 m could possibly bea thin annual layer that cannot be fully resolved, but mayalso just arise from unusually shaped annual peaks.1232 annual layers are marked in the Younger Dryas, ofwhich 78 are uncertain. The uncertain layers fall in two cat-egories: layers that are only supported by evidence in someof the series (type I), and as the one illustrated, shoulders,wide peaks, or double peaks that could represent two thinannual layers not fully resolved or one annual layer repre-sented by peaks with unusual shapes (type II). Type II layersaccount for almost 3/4 of the uncertain layers in the YoungerDryas.4.2.3. Bølling and Allerød Sections, Depth 1526.52–1604.64 m

The Bølling and Allerød sections proved to be the mostchallenging section to date, due to the very changeable na-ture of the data (and climate) in this time interval. Stableisotope data show that climatic conditions generally changefrom a rather warm climate at the beginning of the Bølling toa much cooler climate at the end of the Allerød, but also thatthe temperature changes abruptly towards both cooler andwarmer conditions several times during this period. Thisvariability is clearly observed in all data series. The season-ality of the series changes rapidly several times; from condi-tions similar to those in the Younger Dryas, where all seriespeak simultaneously, to Holocene-like conditions where theseries have different seasonality. These changes do not al-ways happen synchronously with changes in the concentra-tion levels, isotopic values, or observed annual layer thick-ness. A detailed study of the different timing of the changesin the different data series can increase the understandingof the physical processes governing the climate system, butthis is beyond the scope of this work. However, the observedchanges of the properties of the data make the identificationof annual layers difficult, even when all data series are avail-able.In the warmer parts of Bølling and Allerød, the [NH+

4 ] series

X - 8 RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY

1536.9 1537 1537.1 1537.2 1537.3Depth (m)

020406080

[NH

4+ ] (pp

bw)

255075100125

[Ca

2+] (ppbw)

0

100

200

[NO

3− ] (pp

bw)

0

50

100

[Na

+] (ppbw)

0

100

200

[SO

42−] (

ppbw

)

0.5

1

1.5

2

Cond. (µS

/cm)

10−4

10−2

100

EC

M 0

50

100

150

Dust (part./µL)

406080

100V

is. S

trat

.

Figure 6. Example of data (heavy lines), resolution enhanced data (thin lines), and annual layer mark-ings (grey bars) from a cold period in Allerød. In this section the [NH+

4 ] and [NO−

3 ] series do not showclear annual cycles, while the other series peak almost simultaneously. The open grey bars indicateuncertain layers. The annual layer marks have been set in the annual peaks of [Ca2+].

proves useful for dating, as observed in the Holocene. TheVS and conductivity series show clear annual cycles in thesections where most series peak simultaneously. The [NH+

4 ]and [NO−

3 ] series are observed to have a peculiar tendency tocontain additional simultaneous summer peaks not presentin the other series, a phenomenon not encountered elsewhere

in the data used in this work. In Figure 6, a section from arelatively cold part of Allerød is shown. It is apparent thatthe relative timing of the different series is less constantthan in the Holocene and Younger Dryas. Note also howthe [NH+

4 ] and [NO−

3 ] series have roughly the same num-ber of peaks as the other series, but without exhibiting a

1572.7 1572.8 1572.9 1573 1573.1 1573.2Depth (m)

0

100

200

[Ca2+

] (pp

bw)

0

100

200

[NO

3 −] (ppbw)

0

50

100

150

[Na+ ] (

ppbw

)

0

100

200

300

[SO

4 2−] (ppbw)

0

2

4

Con

d. (µ

S/c

m)

10−4

10−3

10−2

EC

M

406080

100

Vis

. Str

at.

Figure 7. Example of data (heavy lines), resolution enhanced data (thin lines), and annual layer mark-ings from the cold period between Bølling and Allerød (MIS-1d), with δ18O values close to that of thesection shown in Figure 6. However in this section, all series peak simultaneously, and there are novariations in seasonality. The [NO−

3 ] and ECM series show clear annual cycles, and the annual layerscould be identified with reasonable certainty based on the ECM and VS series alone. The [NH+

4 ] signalis not shown, as it does not exhibit a clear annual signal.

RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY X - 9

clear annual cycle, and that the dust mass series only barelyresolves the annual cycle. The ECM does not show clearannual cycles, and there are substantially more peaks in VSthan those associated with the annual layers. A section atthe boundary between Bølling and Allerød (inside MIS-1d)is shown on Figure 7. It is also a relatively cold period,having δ18O-values only slightly higher than those in theAllerød section in Figure 6. The dust and [NH+

4 ] series havebeen excluded, as they do not show annual cycles, while the[NO−

3 ] and ECM series again are showing clear annual cy-cles. Note also how the series peak almost simultaneously,and that the annual layers can be placed based almost onthe ECM and VS series alone. The sections in Figures 6and 7 would be expected to be rather similar from their iso-topic values alone, but the differences clearly illustrate thechallenges of identifying annual layers in periods with highlyvariable climatic conditions.Of the 1843 annual layer markings in the Bølling andAllerød, 94 are uncertain layers. About 10 are placed tomark layers that are uncertain because of data quality prob-lems and data gaps, and the rest are equally divided betweenthe two types defined in section 4.2.2.4.2.4. Oldest Dryas Section, Depth 1604.64 m andbelow

In the Oldest Dryas, the phasing of the series is simi-lar to the phasing observed in the Younger Dryas. Whencomparing the data resolution with estimated model annuallayer thicknesses from the ss09sea model of NGRIP members[2004], it is apparent that the resolution is marginal in theOldest Dryas. An inspection shows that the data series doindeed contain many double peaks, wide peaks, and shoul-ders suspected to contain additional annual layers. As theinitial approach, we use the same criteria as in the YoungerDryas, marking the most prominent of these features withtwo annual layer markings and the less pronounced with anannual layer marking plus a type II uncertain layer mark-ing. However, it is clear that many weak indications remainin the data that could possibly represent additional annuallayers. Because of the marginal resolution of the CFA data,more emphasis has to be put on the highly resolved VS andECM data series, and only the last few meters of the Old-est Dryas before the transition into Bølling have thereforebeen included in the presented time scale. The time scalefor these few meters is slightly less reliable, but because ofthe shortness of the section (87 marks, of which 7 are un-certain), and the fact that the isotope values (and thus theexpected annual layer thickness) do not attain full glacialvalues, the problem is of minor importance.

5. Constructing the Final Time Scale

As described in section 4, the datum of the presentedtime scale is the volcanic horizon inside the 8.2 ka cold eventdated in the DYE-3 ice core to 8236 b2k with a maximumcounting error of 47 years. In the Holocene I section, therevised time scale based on GRIP CFA data is used. TheGRIP time scale has been transferred to NGRIP by match-ing the ECM signals of GRIP and NGRIP. The GRIP [NH+

4 ]series contains clear annual peaks, and because peaks in the[NH+

4 ] series lead to minima in the ECM signal, it is possibleto match the two series on an annual scale. The matchingcan be ambiguous, but for every few tens of years better fixpoints are supplied by strong ECM peaks, primarily relatedto volcanism, or by recognizable patterns in the [NH+

4 ] andECM curves. The mismatch is estimated to be maximumone year at fix points and maximum 3 years in long sectionswithout fix points. The result is a common Holocene I timescale for GRIP and NGRIP starting with the year 7903 b2k(GRIP depth 1299.82 m, NGRIP depth 1194.05 m), leading

up to the year 10,276 b2k (GRIP depth 1522.79 m, NGRIPdepth 1404.71 m) after which the NGRIP CFA data start.In the Holocene II section, the four-investigator time scalebased on NGRIP data and the two-investigator time scalebased on both GRIP and NGRIP data have been combined.The latter is constructed without uncertain layer markings,while the first one contains 23 uncertain marks. On top ofthe uncertain layers only present in the 4-investigator timescale, the number of years found in one but not the othertime scale amounts to 9 and 7 annual layer markings, re-spectively. In order to reach a common result, the two timescales were compared and combined by one investigator fromeach group. It was agreed how to interpret the combinedGRIP and NGRIP data in each of the 39 points of disagree-ment. The use of uncertain layer markings has been adoptedfor this combined time scale, and the uncertain layers werecounted as 0.5 year in the construction of the final timescale. The resulting Holocene II time scale is valid for bothcores, covering the age interval 10,277–11,703 b2k, and con-tains 6.6% more annual layers than the existing GRIP timescale [Johnsen et al., 1992]. Finally, the time scale of theHolocene I and II periods was transferred to DYE-3 depthsusing the same ECM-based matching procedure, making theGICC05 time scale valid for all three cores.The annual layer counting in NGRIP continues back to adepth of 1607 m, reaching a few meters below the OldestDryas – Bølling transition, which is found at a depth of1604.64 m, corresponding to an age of 14,692 b2k. As above,the uncertain marks have been counted as 0.5 year.Below the YDPB transition, matching of the GRIP andNGRIP series becomes more difficult as the GRIP CFA failsto resolve the annuals in the Younger Dryas, and the annualcycles in the ECM signal become hard to identify. The timescale obtained from the NGRIP CFA data can therefore notreadily be transferred on a year-by-year basis to the GRIPcore below the YDPB transition. Work by Seierstad et al.[2006] focuses on the possibility of matching the GRIP andNGRIP records at greater depths, and on identifying annuallayers based on high-resolution δD data within the Bølling-Allerød period in the GRIP core, and future progress in thiswork will hopefully extend the time range where GRIP datais available on the GICC05 time scale.

5.1. Assessing the Uncertainty

Contributions to the uncertainty of a stratigraphic icecore time scale include problems with the core stratigraphy,core loss during drilling and handling, data loss during sam-pling and measurements, insufficient measuring resolution,and misinterpretation of the annual layer record [Alley etal., 1997].5.1.1. Uncertainty from Imperfect Core Stratigraphy

The basic assumption that the ice core comprises anunbroken sequence of precipitation from the past can beerroneous due to missing precipitation or due to post-depositional processes like re-deposition or melting. Therelatively high accumulation rates and small surface slopesmake loss of full years due to missing precipitation or windscouring unlikely [Fisher et al., 1983], and melt events areknown to occur only extremely rarely at the GRIP andNGRIP drill sites. In the 1400 year long Holocene II sectionwhere both GRIP and NGRIP CFA data exist, the recordswere matched year to year, and no indication of missingyears was found when comparing the records, substantiat-ing the assumption of the ice cores being unbroken annuallayer sequences. This is also supported by the discussion insection 6. In cold periods with lower accumulation, and pos-sibly more stormy conditions, this may not be true, but theuncertainty contribution from imperfect core stratigraphy isconsidered to be insignificant in the time periods spannedby this work.5.1.2. Uncertainty from Data Gaps

X - 10 RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY

In the depth interval considered in this work, core loss isnonexistent, and only very short sections of the NGRIP corecould not be sampled for the CFA measurements. Aroundirregular core breaks, the ends of the CFA sample pieceshad to be trimmed, resulting in a few centimeters of missingCFA data, but VS and ECM measurements are often avail-able across these short CFA data gaps. Most of the gapsare less than a centimeter long, not causing ambiguities inthe process of annual layer identification. Longer data gapsof 2–10 cm (corresponding to about once or twice the meanannual layer thickness) occur on average every five meters orso, but using the ECM and VS data, it is usually possible toplace the annual layer markings without significant uncer-tainty. In addition to the short sections where all CFA dataare missing, one or more of the CFA series are sometimesmissing over longer intervals due to problems with one ormore of the analysis subsystems. In the cases where annuallayer identification has been difficult due to missing data se-ries, the uncertainty has been indicated by using uncertainlayer marks. The total uncertainty due to missing data isthus estimated to be around ten years out of the 4000–5000years spanned by the NGRIP data in this work, and mostof the uncertainty is accounted for by the use of uncertainlayer markings.In the GRIP Holocene I section there are occasional CFAdata gaps affecting one or two of the series, totalling about10 m, plus a number of smaller gaps which do not causeproblems for the annual layer identification. Most of thelonger gaps contain 2–4 annual layer marks, but the uncer-tainty contribution is small because all species are not miss-ing at the same time, and because ECM data are availableacross the CFA data gaps. The estimated total uncertaintycontribution from data gaps is a few per mille, of which themajor part is accounted for via the use of uncertain layermarks.5.1.3. Uncertainty from Insufficient Measuring Res-olution

The combined CFA, ECM, isotope, and VS data set isbelieved to resolve the annual layers well in the Holoceneand in the warm parts of the Bølling and Allerød, while ex-traordinarily thin annual layers may show up only faintly orbe lost in the Younger Dryas and in the coldest parts of theAllerød period.In the Holocene section, only a couple of features that couldrepresent poorly resolved layers were identified. These weremarked with uncertain layer marks. It is not considered tobe likely that a significant number of years have vanishedcompletely due to insufficient resolution.Double peaks suspected to contain more than one year aremuch more common in the Younger Dryas section. In addi-tion, a small number of years may have been lost altogetherdue to insufficient resolution at the end of Allerød and in theYounger Dryas. Special care was taken not to miss thin lay-ers represented as poorly separated double peaks or shoul-ders on neighbouring peaks. Depending on the amount ofevidence present, these features were marked as two layersor as a layer and an uncertain layer. The upper limit of thenumber of lost layers is estimated to be a few tens of yearsin the section comprising the Younger Dryas and the end ofAllerød, corresponding to about 1% of the section’s annuallayers.5.1.4. Uncertainty from Erroneous Interpretation ofthe Annual Layer Record

The vast majority of annual layers stand out clearly in alldata series, but uncertainty is introduced when an annuallayer is backed up by evidence only in some of the data series,or when a certain well-resolved feature is suspected to con-tain more than one annual layer. Also, the basic assumptionthat the annual layers in ice cores are represented by peaks

in almost all the individual data series may not be perfectlytrue. This uncertainty contribution is the dominant partof the total uncertainty, but can only be partially assessedquantitatively. The cases of ambiguity in the annual layeridentification process have been identified using the uncer-tain layer markings, but in addition one must expect thatthere can be a bias in the annual layer identification process.This is clearly illustrated by the result of the revision of theGRIP Holocene time scale. As described in section 4.1 and5, the number of annual layers in the Holocene I and II sec-tions have increased by 1.5% and 6.6% from the old to thenew GRIP time scale, respectively. These differences reflecta change in the way data are interpreted, rather than count-ing errors, which is reflected by the fact that the increasein the number of annual marks in the Holocene II section is4–5 times larger than the number of uncertain layer marksin the same section.

The comparison of the 2-investigator and the 4-investigator time scales of the Holocene II section consti-tutes the only place where we can make an independentbias magnitude estimate. The two time scales were madeusing partially the same data, but the two groups of investi-gators worked independently of each other. The differencesare described in section 5, and indicate a bias level of 1%or less in the Holocene II section. Below the YDPB transi-tion, we do not have the opportunity of checking the timescale with independent ice core data, and thus only aimat estimating the maximum counting error, which we de-rive from the number of uncertain layer marks. Over thetotal depth interval in question here, 294 of 7021 annuallayer marks (corresponding to 4.2%) were of the uncertaintype. As the uncertain layer marks are regarded as 0.5±0.5year in the final time scale, the number of years in the sec-tion is 6874 with a maximum counting error of 147 yearsassuming that the counting errors are correlated. In Ta-ble 3, the number of layer markings within each climatic

1200 1300 1400 1500 16005

10

15

20

25

λ corr

. (cm

)

Depth (m)

Observed+ uncertainty band

Figure 8. Observed annual accumulation rates afterstrain correction, λcorr., averaged over 2 meter sections.The shaded uncertainty band is derived from the uncer-tain annual layer marks (in the Holocene II and glacialparts), and thus represents only the part of the uncer-tainty that is related to ambiguous features in the dataset. In the Holocene I section, a constant 2% uncer-tainty band is used. The high variability of λcorr. in theHolocene I section is likely to be caused by a combinationof two effects: (1) at smaller depths each 2 meter intervalcontains fewer annual layers resulting in increased scatteraround the mean value, and (2) the transfer of the GRIPdata based time scale to NGRIP depths using commonECM fix points introduces an uncertainty in the numberof annual layers in each 2 meter interval, and thus in thederived accumulation rates, although the total number ofannual layers remains accurate. The accumulation ratesare sensitive to how the strain correction is performed,and should be regarded as preliminary results.

RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY X - 11

Table 3. Estimated maximum counting error excluding possible bias in the annual layer identification processfor the different climatic periods.

Section NGRIP Number of annual layers Duration Maximum counting errordepth (m) Certain Uncertain (y) absolute (y) relative

Holocene I 1194.05–1404.74a 2326 96 2374 48 2.0%Holocene II 1404.74–1492.45 1418 19 1427 10 0.67%Younger Dryas 1492.45–1526.52 1154 78 1193 39 3.3%Allerød 1526.52–1574.80 1147 61 1178 31 2.6%Bølling 1574.80–1604.64 602 33 618 17 2.7%End of Oldest Dryas 1604.64–1607.00 80 7 84 4 ∼ 4%Total 6727 294 6874 147 2.1%

a Time scale based on GRIP CFA data, corresponding GRIP depths: 1299.81–1522.75 m.

period are listed together with the maximum counting er-ror derived from the number of uncertain layer markingsin this way. The counting error derived from the uncer-tain layer markings heavily depends on whether the uncer-tainties are assumed to be correlated or uncorrelated, andconsequently on which error summation procedure shouldbe applied. If the counting errors are assumed to be fullyuncorrelated, the errors should be summed in a quadraticsense, producing a maximum counting error estimate of√

(

1

2

)2+ . . . +

(

1

2

)2=

√

294(

1

2

)2=

√

294

2≈ 9. Using the

more reasonable assumption that the errors are fully corre-lated within each climatic period and uncorrelated from oneperiod to the other, the maximum error estimate becomes√

482 + 102 + 392 + 312 + 172 + 42 = 72, using the perioddivision from Figure 1. This approach was also used whenderiving counting error estimates for the previous GRIPtime scale, but has the obvious disadvantage that the maxi-mum counting error estimate depends on the choice of which

0.4 0.6 0.8 1 1.2 1.40

0.02

0.04

0.06

0.08

log10

(λcorr.

/cm)

Obs

erve

d fr

eque

ncy

YD Bølling Holocene IIDepth : 1495−1523 1576−1603 1404−1490

µ : 0.790 1.064 1.121σ : 0.124 0.121 0.090

10µ : 6.2 cm 11.6 cm 13.2 cm

3 5 7 10 20 30

1

2

3

4

λcorr.

(cm)

Probability density

Figure 9. Distribution of strain corrected annual layerthicknesses λcorr in the Holocene II, Younger Dryasand Bølling periods (thin lines). The distributions oflog(λcorr) in each of the three periods are close to Gaus-sian distributions (thick lines). The coefficients of thefitted Gaussian curves are listed. The annual layer thick-ness is thus seen to have a log-normal distribution, andthe chances of encountering an annual layer with halfand double the mean thickness, respectively, are thusequal (see section 6). The mean accumulation rates (the10µ values) derived from the fitted curves are sensitiveto how the strain correction is performed, and should beregarded as preliminary results.

periods are considered independent. If for example theAllerød period is considered to consist of 4 independent pe-riods (corresponding to the GIS1a–e of Bjorck et al. [1998]),the 61 uncertain layer marks only result in a maximum

counting estimate of

√

(

9

2

)2+

(

10

2

)2+

(

38

2

)2+

(

4

2

)2 ≈ 20

years instead of 31 years if Allerød is considered to be oneperiod. Recognizing that the counting errors in reality areneither uncorrelated nor fully correlated, we adopt the sim-ple and conservative approach, summing up the uncertain-ties as if they were correlated. The counting error estimatespresented here are thus highly conservative, and as the max-imum counting error dominates over the sources of uncer-tainty associated with imperfect stratigraphy and data gapsby about an order of magnitude, only the maximum count-ing error is given. The uncertainty from bias in the annuallayer identification process can by the nature of the prob-lem not be estimated without the existence of independentdata sources, and is thus not included in the error estimatespresented.

6. Distribution of Annual Layer Thicknesses

Figure 8 shows the annual layer thickness profile derivedfrom the NGRIP core using the GICC05 time scale. Theuncertainty band indicated in the figure is derived from theuncertain layer marks, and thus represents the part of theuncertainty that arise from features in the data that the in-vestigators found ambiguous. The strain correction has beenperformed using a first order flow model, where the strainǫ at depth z is given as ǫ = 1 − z

z0. The value z0 = 2680

m is used, whereby ǫ closely resembles the strain historyderived from the ss09sea model [NGRIP members, 2004] inthe 1200–1600 m depth interval. As the ss09sea model timescale is significantly different from the GICC05 time scale inthis interval, the absolute accumulation rate values shouldtherefore be regarded as preliminary. However, the ratiosbetween the accumulation rates in the Holocene II, YoungerDryas, and Bølling periods are robust to a wide range ofreasonable z0 values.Distributions of strain corrected observed annual layer thick-nesses within three selected periods are shown in Figure 9(note the logarithmic scale). To ensure that the climaticconditions are rather constant within each period, the vari-able Allerød period has been excluded, and a few meters ateach end of the periods have been removed to ensure that thetransitions between the individual periods do not influencethe distributions. It is apparent from the figure that theannual layer thicknesses to a good approximation are log-normally distributed, and that the variability of the annuallayer thicknesses is by far smallest in the Holocene II section,and roughly identical for the Younger Dryas and Bøllingsections. If the conclusion that annual layer thicknesses arelog-normally distributed holds in general, mean and stan-dard deviation values of accumulation rates should be cal-culated from logarithmic transformed data rather than from

X - 12 RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY

Table 4. Age and maximum counting error estimates for selected horizonsa

Horizon NGRIP Age Total maximumdepth (m) (b2k) counting error (y)

Upper limit of presented time scale 1194.05 7,903 41Time scale datum (8.2 ka cold event) 1228.24 8,236 47Saksunarvatn volcanic layer 1409.83 10,347 89YD – Preboreal transition 1492.45 11,703 99Vedde volcanic layer (Z1) 1506.14 12,171 114Onset of Younger Dryas 1526.52 12,896 138End of Bølling 1574.80 14,075 169Onset Bølling 1604.64 14,692 186Lower limit of presented time scale 1606.96 14,776 190

a The total maximum counting error consists of a maximum counting error derived as described in section 5.1.4 plus the 41 yearmaximum counting error from the dating of the upper 7.9 ka in the DYE-3 ice core, but does not include possible bias in the annuallayer identification process. All ages are reported relative to A.D. 2000 (b2k).

the observed annual layer thicknesses, contrary to commonusage, because the notion of standard deviation makes moresense when calculated for data that are approximately sym-metrically distributed. Using the mean accumulation ratevalues derived from the fitted log-normal curves of Figure9, the accumulation rate in the Younger Dryas and Bøllingperiods are 47% and 88% of the Holocene II value, respec-tively, or 48% and 89% of the Holocene II values if ordi-nary mean values are used. These ratios are robust within±1% for z0 ∈ [2580, 2780]m. Although the strain profile de-rived from the ss09sea model is not fully consistent with theGICC05 time scale, the inaccuracy of the strain profile isregarded to be smaller than the effect of varying z0 withinthe bounds described, and we thus regard the accumulationrate ratios presented to be precise at least within ±2%. Theratios indicate a smaller contrast between the stadial andinterstadial accumulation rates than those observed in theGISP2 ice core where the similar ratios are approximately43% and 97% [Alley et al., 1993; Cuffey and Clow, 1997].Being log-normally distributed, annual layers with e.g. dou-ble and half the mean annual thickness λmean, respectively,will occur with equal probabilities. Using the fitted distribu-tions, the probability of a random annual layer being eitherthicker than 2λmean or thinner than λmean/2 can be esti-mated to be below 0.1% for the Holocene II section, 1.5%for the Younger Dryas, and 1.3% for Bølling. When com-pared with the resolution of the data used in this work, thedistribution of the annual layer thicknesses for the HoloceneII section indicates that it is extremely unlikely that anyannual layers have been missed in the Holocene parts of thiswork due to insufficient resolution, and that the problemalso should be negligible in the Bølling section.

7. Discussion and Conclusions

The GICC05 time scale across the last termination (7,903– 14,776 b2k) has been constructed by identifying and count-ing annual layers using multi-parameter data sets from theGRIP and NGRIP ice cores. The ages of the onset and endof the Younger Dryas, Bølling and Allerød periods can befound in Table 4 along with the ages of the Saksunarvatnand Vedde volcanic ash layers. In Figure 10, 20 year meanvalues of NGRIP δ18O data are shown on the GICC05 timescale. For comparison, the same δ18O data are shown on thepreviously used time scales: the existing counted time scalefor the Holocene part and the ss09sea model time scale be-low the YDPB transistion [Johnsen et al., 1992; 2001]. TheGRIP isotope profile is not displayed, as the ss09sea timescale is common to the GRIP and NGRIP cores [NGRIPmembers, 2004]. The 20 year resolution GISP2 isotope pro-file of Grootes and Stuiver [1997] and Stuiver and Grootes[2000] is presented on the time scale of Meese et al. [1997].The differences between the time scales are apparent in Fig-ure 11, where the differences in the dating of 46 selectedECM horizons in the three time scales are shown.

At the datum of the GICC05 time scale (8236 b2k), the ex-isting NGRIP time scales yield ages that are about 20 yearsyounger than the GICC05, but the difference grows increas-ingly fast towards the YDPB transition. The transition hasbeen dated to 11,703 b2k with a maximum counting errorof 99 years, pushing the YDPB transition about 150 yearsback relative to the previous GRIP and NGRIP age esti-mates. The difference increases slowly to about 180 years at12.5 ka b2k, and then monotonically decreases again throughthe Allerød and Bølling periods until the onset of Bølling,where the difference has been reduced to about 50 years.The differences between the former and new time scales inthe Holocene part reflect the new interpretation of the GRIPdata, while the differences below the YDPB transition in-dicate that the relationship between δ18O and accumula-tion used to construct the model time scales [Johnsen et al.,1995; Dahl-Jensen et al., 1993; Johnsen et al., 2001] needsimprovement in the Bølling and Allerød periods.In general, the GICC05 time scale agrees better with theGISP2 time scale than the former GRIP and NGRIP timescales. At the 8.2 ka event the GISP2 time scale yields datesthat are 36 years older than the corresponding GICC05 ages,but the main difference between the GICC05 and GISP2time scales in the Holocene is the number of annual layersin the 8.2–9.5 ka b2k section, where the GISP2 time scalelacks about 60 years, or 5%, relative to the GICC05. Itshould be noted that the investigators producing the GISP2time scale did not agree on the number of years in the GISP21371 – 1519 m depth interval (GISP2 age 8070 - 9424 b2k),where R.B. Alley counted 72 years more than the numberof years in the official GISP2 time scale after the ice corehad been stored for a few years [Alley et al., 1997, Table2]. The GICC05 and GISP2 time scales have roughly thesame number of years in the 9.5–11.5 ka b2k section, andagree within a few years on the age of the YDPB transitionwhen the transition depth is defined using deuterium excessdata [T. Popp, pers. comm., 2005]. However, the differencegrows rapidly in the Younger Dryas section. In the 11.5–12.9 ka b2k section, corresponding to the Younger Dryasand the first 2 centuries of the Preboreal, the GISP2 timescale contains 84 years, or 6%, more years than the GICC05.In the Allerød the two scales agree fairly well again, whileGISP2 has about 40, or 6%, years less in the Bølling periodrelative to the GICC05. Significant differences which cannotbe attributed to counting uncertainty thus remain betweenthe GICC05 and GISP2 time scales.As previously mentioned, the GICC05 time scale has notbeen transferred to the GRIP depth scale on a year-to-yearbasis below the YDPB transition due to marginal data res-olution, but common features in the ECM signal allow formatching of the two cores. Interpolation is used between fixpoints, leading to an estimated maximum matching error of5 years. Bjorck et al. [1998] defines the Younger Dryas asthe GRIP depth interval 1623.6 – 1661.5 m and gives theduration as 1150±50 years. The matching of the GRIP andNGRIP ECM records makes it possible to identify the cor-responding depth interval in the NGRIP core, and according

RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY X - 13

8.0

8.5

9.0

9.5

10.0

10.5

11.0

11.5

12.0

12.5

13.0

13.5

14.0

14.5

15.0

−43 −41 −39 −37 −35 −33

NGRIP δ18O (per mille)

previous time scales

NG

RIP

1 data(old counted tim

e scale)N

GR

IP2 data

(old counted time scale)

NG

RIP

2 data (ss09sea model)

−43 −41 −39 −37 −35 −33

NGRIP δ18O (per mille)

new GICC05 time scale

−43 −41 −39 −37 −35 −33

GISP2 δ18O (per mille)

Meese et al. time scale

NG

RIP

1 dataN

GR

IP2 data

Age(kyr B2k)

Datum :8236 b2k

Figure 10. caption in next column

Figure 10. Stable isotope profiles from NGRIP andGISP2 of the entire sections covered by the presented timescale. The red curve shows 20 year mean values of NGRIPδ18O data on the existing counted time scale (Holocene part)and ss09sea model time scale (glacial part). The blue curveshows the same data on the new GICC05 time scale. Thedifference between the shape of the blue and red curves is aconsequence of different 20 year averaging intervals, as theunderlying δ18O data are the same. The green curve showGISP2 δ18O data on the time scale of Meese et al. [1997].The black bullets to the right are the fix points used for thecomparison in Figure 11, shown relative to the GISP2 curve.−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−

to GICC05 the duration of the Younger Dryas is evaluatedto be 1186 years (maximum counting error 44 years). Thedeviation from the duration given in Table 3 arises from thedifferent definitions of onset and end depths. The GICC05timescale thus agrees well with the INTIMATE duration es-timate of Bjorck et al. [1998], even though the onset andtermination of the Younger Dryas are 150-200 years olderaccording to GICC05. In the same way, the total dura-tion of the Bølling – Allerød period (GIS-1a through 1e)is given as 2050±50 years by Bjorck et al. [1998] (GRIPdepth 1661.5 – 1753.4 m), while the corresponding durationis only 1818 (maximum counting error 53 years) accordingto the GICC05 time scale. The combination of the Bølling– Allerød being 200-250 years shorter and the YDPB tran-sition being 150 years older in the GICC05 compared toBjorck et al. [1998], means that the age of 14,750±50 b2kfor the onset of Bølling given by Bjorck et al. [1998] agreesfairly well with the GICC05 age of 14,692 b2k (maximumcounting error 186 years).It should be emphasized that the maximum counting errorsgiven here reflect a conservative estimate of the maximumerror associated with interpretation of ambiguous features inthe data, data gaps, and marginal resolution in accordancewith the discussion in section 5.1.4, but does not includeuncertainty contributions from possible bias in the annuallayer identification process, that can not be quantitativelyassessed without independent data.The relative phasing of the different impurity data seriesis observed to be different during cold and warm condi-tions, indicating that the annual distribution of precipita-tion changes rapidly both at the climate transitions andwithin the Bølling and Allerød periods. The annual layerthicknesses are observed to be log-normally distributed withgood approximation, and the ratios of the mean accumula-tion rates of Younger Dryas and Bølling to that of the EarlyHolocene are 47 ± 2% and 88 ± 2%, respectively.The work with the new Greenland Ice Core Chronology con-tinues, and the time scale presented here will be both ex-tended further back in time and compared and validated bycomparison with results from independent dating strategies.Only by providing the ice cores with reliable time scales, thefull value of the records extracted from the ice cores can beappreciated and used in conjunction with other palaeocli-matic data, thereby assessing essential questions about thetiming of past climatic changes.

X - 14 RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY

8. Online data access

20 year mean values of GRIP and NGRIP δ18O data onthe GICC05 time scale (as shown for NGRIP in Figure 10)can be downloaded from http://www.icecores.dk.

9 10 11 12 13 14

−80

−40

0

40

80

120

160

GICC05 age (ka b2k)

GIC

C05

−G

ISP

2 ag

e di

ffere

nce

GIC

C05

−ex

istin

g N

GR

IP a

ge d

iffer

ence

Holocene I Holocene II YD Allerød Bøll.

Figure 11. Detailed comparison of the GICC05 timescale with the existing NGRIP/GRIP time scales (spec-ified in the caption of Figure 10) and the GISP2 timescale of Meese et al. [1997] using the dates of 46 com-mon ECM events in the NGRIP, GRIP, and GISP2 cores(see Figure 10). Positive values indicate that an event isoldest according to the GICC05 time scale.

Acknowledgments. This work is a contribution of theCopenhagen Ice Core Dating Initiative which is supported by agrant from the Carlsberg Foundation.NGRIP is directed and organized by the Ice and Climate ResearchGroup at the Niels Bohr Institute, University of Copenhagen,Denmark. It is supported by funding agencies in Denmark (SNF),Belgium (FNRS-CFB), France (IPEV and INSU/CNRS), Ger-many (AWI), Iceland (RannIs), Japan (MEXT), Sweden (SPRS),Switzerland (SNF) and the USA (NSF, Office of Polar Programs).SOR gratefully acknowledges Robert Mulvaney, and BritishAntarctic Survey, for support during a visit to BAS from Au-gust to October 2004.We thank Eric Wolff for comments during review that greatlyimproved the quality of the uncertainty discussion.

References

Alley, R., et al. (1993), Abrupt increase in Greenland snow ac-cumulation at the end of the Younger Dryas event, Nature,362 (6420), 527–529.

Alley, R. B., et al. (1997), Visual-stratigraphic dating of theGISP2 ice core: Basic, reproducibility, and application, Jour-nal of Geophysical Research, 102 (C12), 26,367–26,381.

Anklin, M., R. C. Bales, E. Mosley-Thompson, and K. Steffen(1998), Annual accumulation at two sites in Northwest Green-land during recent centuries, Journal of Geophysical Research,103 (D22), 28,775–28,783.

Beer, J., et al. (1991), Seasonal variations in the concentrations of10Be, Cl−, NO−

3, SO2−

4, H2O2, 210Pb, 3H, mineral dust, and

δ18O in Greenland snow, Atmospheric Environment, 25 (19),899–904.

Bigler, M. (2004), Hochauflosende Spurenstoffmessungen anpolaren Eisbohrkernen: Glaziochemische und klimatischeProzessstudien, Ph.D. dissertation, University of Bern,Switzerland.

Bjorck, S., M. J. C. Walker, L. C. Cwynar, S. Johnsen, K.-L.Knudsen, J. J. Lowe, B. Wohlfarth, and INTIMATE Members(1998), An event stratigraphy for the Last Termination in theNorth Atlantic region based on the Greenland ice-core record:

a proposal by the INTIMATE group, Journal of QuaternaryScience, 13 (4), 283–292.

Blunier, T., et al. (1998), Asynchrony of Antarctic and Green-land climate change during the last glacial period, Nature, 394,739–743.

Bond, G., W. Broecker, S. Johnsen, J. McManus, L. Labeyrie,J. Jouzel, and G. Bonani (1993), Correlations between climaterecords from North Atlantic sediments and Greenland ice, Na-ture, 365 (6442), 143–147.

Bory, A.-M., P. Biscaye, A. Svensson, and F. Grousset (2002),Seasonal variability in the origin of recent atmospheric mineraldust at NorthGRIP, Greenland, Earth and Planetary ScienceLetters, 196, 123–134.

Clausen, H., C. Hammer, C. Hvidberg, D. Dahl-Jensen, J. Stef-fensen, J. Kipfstuhl, and M. Legrand (1997), A comparison ofthe volcanic records over the past 4000 years from the Green-land Ice Core Project and Dye3 Greenland ice cores, Journalof Geophysical Research, 102 (C12), 26,707–26,723.

Cuffey, K., and G. Clow. (1997), Temperature, accumulation,and ice sheet elevation in central Greenland through the lastdeglacial transition, Journal of Geophysical Research, 102,26,383–26,396.

Dahl-Jensen, D., S. J. Johnsen, C. U. Hammer, H. B. Clausen,and J. Jouzel (1993), Past accumulation rates derived fromobserved annual layers in the GRIP ice core from Summit,Central Greenland, in Ice in the Climate System, NATO ASISer. I, vol. 12, edited by R. W. Peltier, pp. 517–532, Springer-Verlag, New York.

Dahl-Jensen, D., N. S. Gundestrup, H. Miller, O. Watanabe, S. J.Johnsen, J. P. Steffensen, H. B. Clausen, A. Svensson, andL. B. Larsen (2002), The NorthGRIP deep drilling programme,Annals of Glaciology, 35, 1–4.

Dansgaard, W., et al. (1993), Evidence for general instability ofpast climate from a 250-kyr ice-core record, Nature, 364 (6434),218–220.

Fischer, H., and D. Wagenbach (1996), Large-scale spatial trendsin recent firn chemistry along an east-west transect throughcentral Greenland, Atmospheric Environment, 30 (19), 3227–3238.

Fisher, D., R. Koerner, W. Paterson, W. Dansgaard, N. Gunde-strup, and N. Reeh (1983), Effect of wind scouring on cli-mate records from ice-core oxygen-isotope profiles, Nature,301, 205–209.

Fuhrer, K., A. Neftel, M. Anklin, and V. Maggi (1993), Contin-uous measurements of hydrogen peroxide, formaldehyde, cal-cium and ammonium concentrations along the new GRIP icecore from Summit, Central Greenland, Atmospheric Environ-ment, 27A(12), 1873–1880.

Fuhrer, K., A. Neftel, M. Anklin, T. Staffelbach, and M. Legrand(1996), High-resolution ammonium ice core record covering acomplete glacial-interglacial cycle, Journal of Geophysical Re-search, 101 (D2), 4147–4164.

Fuhrer, K., E. W. Wolff, and S. J. Johnsen (1999), Timescalesfor dust variability in the Greenland Ice Core Project (GRIP)ice core in the last 100,000 years, Journal of Geophysical Re-search, 104 (D24), 31,043–31,052.

Grootes, P. M., and M. Stuiver (1997), Oxygen 18/16 variabilityin Greenland snow and ice with 10−3 to 105-year time resolu-tion, Journal of Geophysical Research, 102, 26,455–26,470.

Hammer, C. (1989), Dating by physical and chemical seasonalvariations and reference horizons, in Dahlem Konference: TheEnvironmental Record in Glaciers and Ice Sheets, edited byH. Oeschger and J. Langway, C.C., Physical, Chemical, andEarth Sciences Research Report 8, pp. 99–121, John Wiley,New York.

Hammer, C. U., H. B. Clausen, W. Dansgaard, N. Gundestrup,S. J. Johnsen, and N. Reeh (1978), Dating of Greenland icecores by flow models, isotopes, volcanic debris, and continentaldust, Journal of Glaciology, 20 (82), 3–26.

Hammer, C. U., H. B. Clausen, and H. Tauber (1986), Ice-coredating of the Pleistocene/Holocene boundary applied to a cal-ibration of the 14C time scale, Radiocarbon, 28, 284–291.

Hammer, C. U., H. B. Clausen, and C. C. Langway, Jr. (1994),Electrical conductivity method (ECM) stratigraphic dating ofthe Byrd Station ice core, Antarctica, Annals of Glaciology,20, 115–120.

RASMUSSEN ET AL.: A NEW GREENLAND ICE CORE CHRONOLOGY X - 15

Hvidberg, C. S., J. P. Steffensen, H. B. Clausen, H. Shoji, andJ. Kipfstuhl (2002), The NorthGRIP ice-core logging proce-dure: description and evaluation, Annals of Glaciology, 35,5–8.

Johnsen, S., D. Dahl-Jensen, W. Dansgaard, and N. Gundestrup(1995), Greenland palaeotemperatures derived from GRIPbore hole temperature and ice core isotope profiles, Tellus,47B, 624–629.

Johnsen, S. J. (1977), Stable isotope homogenization of polarfirn and ice, in Proc. of Symp. on Isotopes and Impuritiesin Snow and Ice, I.U.G.G. XVI, General Assembly, GrenobleAug./Sept., 1975, pp. 210–219, IAHS-AISH Publ. 118, Wash-ington D.C.

Johnsen, S. J., W. Dansgaard, and J. White (1989), The originof Arctic precipitation under present and glacial conditions,Tellus, 41B, 452–468.

Johnsen, S. J., H. B. Clausen, K. M. Cuffey, G. Hoffmann,J. Schwander, and T. Creyts (2000), Diffusion of stable iso-topes in polar firn and ice: The isotope effect in firn diffusion,in Physics of Ice Core Records, edited by T. Hondoh, pp. 121–140, Hokkaido University Press, Sapporo.

Johnsen, S. J., D. Dahl-Jensen, N. Gundestrup, J. P. Stef-fensen, H. B. Clausen, H. Miller, V. Masson-Delmotte,A. E. Sveinbjornsdottir, and J. White (2001), Oxygen iso-tope and palaeotemperature records from six Greenland ice-core stations: Camp Century, Dye-3, GRIP, GISP2, Ren-land and NorthGRIP, J. Quat. Sci., 16, B01402, doi:10.1029/2003JB002550.

Johnsen, S. J., et al. (1992), Irregular glacial interstadialsrecorded in a new Greenland ice core, Nature, 359, 311–313.

Jouzel, J., and L. Merlivat (1984), Deuterium and oxygen-18 inprecipitation: Modeling of the isotopic effects during snowformation, Journal of Geophysical Research, 89 (D7), 11,749–11,757.

Laj, P., J. M. Palais, and H. Sigurdsson (1992), Changing sourcesof impurities to the Greenland ice sheet over the last 250 years,Atmospheric Enviroment, 26A(14), 2627–2640,.

Langway, Jr., C. C. (1967), Stratigraphic analysis of a deep icecore from Greenland, CRREL Research Report 77, pp. 1–130.

Masson-Delmotte, V., J. Jouzel, A. Landais, M. Stievenard, S. J.Johnsen, J. W. C. White, M. Werner, A. Sveinbjornsdottir,and K. Fuhrer (2005a), GRIP deuterium excess reveals rapidand orbital-scale changes in Greenland moisture origin, Sci-ence, 209, 118–121, doi:10.1126/science.1108575.

Masson-Delmotte, V., et al. (2005b), Holocene climatic changesin Greenland: Different deuterium excess signals at GreenlandIce Core Project (GRIP) and NorthGRIP, Journal of Geophys-ical Research, 110, doi:10.1029/2004JD005575.

Meese, D. A., A. J. Gow, R. B. Alley, G. A. Zielinski, P. M.Grootes, M. Ram, K. C. Taylor, P. A. Mayewski, and J. F.Bolzan (1997), The Greenland Ice Sheet Project 2 depth-agescale: Methods and results, Journal of Geophysical Research,102 (C12), 26,411–26,423.

North Greenland Ice Core Project members (2004), High-resolution record of Northern Hemisphere climate extendinginto the last interglacial period, Nature, 431, 147–151.

Rasmussen, S. O., K. K. Andersen, S. J. Johnsen, M. Bigler,and T. McCormack (2005), Deconvolution-based resolutionenhancement of chemical ice core records obtained by Con-tinuous Flow Analysis, Journal of Geophysical Research,110 (D17304), doi:10.1029/2004JD005717.

Rothlisberger, R., M. Bigler, M. Hutterli, S. Sommer, B. Stauf-fer, H. Junghans, and D. Wagenbach (2000), Technique forcontinuous high-esolution analysis of trace substances in firnand ice cores, Environmental Science and Technology, 34 (2),338–342.

Ruth, U., D. Wagenbach, J. Steffensen, and M. Bigler (2003),Continuous record of microparticle concentration and size dis-tribution in the central Greenland NGRIP ice core duringthe last glacial period, Journal of Geophysical Research., 108,4098, doi:10.1029/2002JD002376.

Seierstad, I. K., S. J. Johnsen, B. M. Vinther, and J. Olsen (2006),The duration of the Bølling-Allerød period (Greenland Inter-stadial 1) in the GRIP ice core, Annals of Glaciology, in press.

Southon, J. (2004), A radiocarbon perspective on Greenland ice-core chronologies: Can we use ice cores for 14C calibration ?,Radiocarbon, 46 (3), 1239–1260.

Steffensen, J. P. (1988), Analysis of the seasonal variation in dust,Cl−, NO−

3, and SO2−

4in two Central Greenland firn cores, An-

nals of Glaciology, 10, 171–177.Stuiver, M., and P. M. Grootes (2000), GISP2 Oxygen

isotope ratios, Quaternary Research, 53, 277–284, doi:10.1006/qres.2000.2127.

Svensson, A., S. W. Nielsen, S. Kipfstuhl, S. J. Johnsen, J. P.Steffensen, M. Bigler, U. Ruth, and R. Rothlisberger (2005),Visual stratigraphy of the North Greenland Ice Core Project(NorthGRIP) ice core during the last glacial period, Journal ofGeophysical Research, 110 (D2), doi:10.1029/2004JD005134.

Taylor, K., et al. (1997), The Holocene-Younger Dryas transitionrecorded at Summit, Greenland, Science, 278, 825–827.

Vinther, B. M., et al. (2006), A synchronized dating of threeGreenland ice cores throughout the Holocene, Journal of Geo-physical Research, in press, doi:10.1029/2005JD006921.

Werner, M., U. Mikolajewicz, M. Heimann, and G. Hoffmann(2000), Borehole versus isotope temperatures on Greenland:Seasonality does matter, Geophysical Research Letters, 27 (5),723–726.

Whitlow, S., P. A. Mayewski, and J. E. Dibb (1992), A com-parison of major chemical species seasonal concentration andacuumulation at the South Pole and Summit, Greenland, At-mospheric Enviroment, 26A(11), 2045–2054.

Sune Olander Rasmussen, Ice and Climate, Niels Bohr Insti-tute, University of Copenhagen, Juliane Maries Vej 30, DK-2100Copenhagen, Denmark. (olander@gfy.ku.dk)